Vehicle Routing Problem with Soft Time Windows Based on Improved Genetic Algorithm for Fruits and Vegetables Distribution

Table 5

Prediction errors of total cost between actual distribution values and predicted values derived from four models during five months.

Twelve months; (times)	Model types
	Improved genetic algorithm		O-X type cross pattern		Group-based pattern		CW pattern
	RMSE		RMSE		RMSE		RMSE
January (25)	0.4838	0.4188	1.7398	1.6216	1.3422	1.2986	1.3652	1.3056
February (26)	0.4759	0.3963	1.7136	1.6163	1.3233	1.2832	1.3469	1.3125
March (15)	0.6563	0.5728	2.0126	1.9023	1.6503	1.3389	1.7027	1.6391
April (13)	0.7229	0.6985	2.2359	2.1362	1.7222	1.6261	1.6918	1.7632
May (11)	0.7935	0.7065	2.4129	2.3061	1.9213	1.8163	2.1356	2.0522
June (12)	0.7368	0.6986	2.3236	2.2346	1.8422	1.7934	1.9589	1.8482
July (10)	0.8159	0.7312	2.6356	2.4995	2.0231	1.9124	2.2677	2.1343
August (12)	0.7263	0.6802	2.2689	2.1693	1.7546	1.5376	1.7703	1.6653
September (17)	0.5421	0.5063	2.0875	1.9574	1.5263	1.4513	1.6896	1.5122
October (21)	0.5116	0.4323	1.8985	1.7023	1.6227	1.5552	1.7263	1.6373
November (22)	0.5032	0.4213	1.8635	1.6986	1.5221	1.4496	1.5894	1.4924
December (23)	0.4978	0.4123	1.7974	1.6629	1.4819	1.3928	1.5431	1.4623